ECOSYSTEM ECOLOGY - ORIGINAL PAPER

Intra-annual variability of anatomical structure and d13C valueswithin tree rings of spruce and pine in alpine, temperateand boreal Europe

Eugene A. Vaganov Æ Ernst-Detlef Schulze ÆMarina V. Skomarkova Æ Alexander Knohl ÆWilli A. Brand Æ Christiane Roscher

Received: 20 August 2007 / Accepted: 13 July 2009 / Published online: 4 August 2009

� The Author(s) 2009. This article is published with open access at Springerlink.com

Abstract Tree-ring width, wood density, anatomical

structure and 13C/12C ratios expressed as d13C-values of

whole wood of Picea abies were investigated for trees

growing in closed canopy forest stands. Samples were col-

lected from the alpine Renon site in North Italy, the lowland

Hainich site in Central Germany and the boreal Flakaliden

site in North Sweden. In addition, Pinus cembra was studied

at the alpine site and Pinus sylvestris at the boreal site. The

density profiles of tree rings were measured using the

DENDRO-2003 densitometer, d13C was measured using

high-resolution laser-ablation-combustion-gas chromatog-

raphy-infra-red mass spectrometry and anatomical charac-

teristics of tree rings (tracheid diameter, cell-wall thickness,

cell-wall area and cell-lumen area) were measured using an

image analyzer. Based on long-term statistics, climatic

variables, such as temperature, precipitation, solar radiation

and vapor pressure deficit, explained\20% of the variation

in tree-ring width and wood density over consecutive years,

while 29–58% of the variation in tree-ring width were

explained by autocorrelation between tree rings. An inten-

sive study of tree rings between 1999 and 2003 revealed that

tree ring width and d13C-values of whole wood were

significantly correlated with length of the growing season,

net radiation and vapor pressure deficit. The d13C-values

were not correlated with precipitation or temperature.

A highly significant correlation was also found between

d13C of the early wood of one year and the late wood of the

previous year, indicating a carry-over effect of the growing

conditions of the previous season on current wood produc-

tion. This latter effect may explain the high autocorrelation

of long-term tree-ring statistics. The pattern, however, was

complex, showing stepwise decreases as well as stepwise

increases in the d13C between late wood and early wood.

The results are interpreted in the context of the biochemistry

of wood formation and its linkage to storage products. It is

clear that the relations between d13C and tree-ring width and

climate are multi-factorial in seasonal climates.

Keywords Dendrochonology � Carbohydrate storage �Picea abies � Pinus cembra � Pinus sylvestris �Tracheid lumen area � Wood density

Introduction

Tree rings are generally considered to be archives of past

climatic conditions and of changes in the environment

(Schweingruber 1996; Wimmer 2002), especially when

trees were growing under extreme conditions in which

changes in climate dominated the response. Despite a long

history of climate reconstructions (Fritts 1966), there is still

some uncertainty regarding the specific effects of certain

climate parameters, especially temperature and drought, on

tree-ring width and wood anatomy (Vaganov et al. 2006).

This uncertainty is based on the knowledge that seasonal

and annual changes in tree-ring anatomy (tree-ring width,

number of cells, size of cells, cell-wall thickness and wood

Communicated by Dan Yakir.

E. A. Vaganov � M. V. Skomarkova

Institute of Forest SB RAS, Akademgorodok,

660036 Krasnoyarsk, Russia

E.-D. Schulze (&) � W. A. Brand � C. Roscher

Max-Planck Institute for Biogeochemistry,

Box 100164, 07701 Jena, Germany

e-mail: dschulze@bgc-jena.mpg.de

A. Knohl

Department of Plant Science, ETH Zurich,

Universitatsstr. 2, 8092 Zurich, Switzerland

123

Oecologia (2009) 161:729–745

DOI 10.1007/s00442-009-1421-y

density) reflect not only processes at the time of xylem-cell

production and maturation, but also additional physiologi-

cal controls and the availability of storage products at the

time of wood formation (Hemming et al. 2001). In addition,

the conditions for wood formation may be different for

solitary growing trees under extreme environmental con-

ditions, which have been generally sampled for tree ring

studies, than for trees growing in forest stands of closed

canopies where competition for water, nutrients and space

in addition to climate be important. Also, single events,

such as late frost or heavy fruiting, may affect wood growth

more than average climatic conditions in forest stands.

In addition to wood anatomy, the isotope composition of

tree rings may serve as an additional parameter to expand

the possibilities of interpreting tree-ring width. The 13C/12C

ratios of wood (cellulose) have also been shown to reflect

climatic conditions (Wilson and Grinsted 1977; Leavitt

1993; McNulty and Swank 1995; Duquesnay et al. 1998;

McCarroll and Loader 2004). The carbon isotope ratios of

carbohydrates are determined by carbon dioxide (CO2)

assimilation and stomatal conductance during the growing

season (Francey and Farquhar 1982; Farquhar et al. 1989;

Brugnoli and Farquhar 2000). However, these ratios in tree

rings may also be influenced by isotopic discrimination

during storage and re-mobilization of storage products,

including various carbohydrates, lipids and proteins, in

leaves (Gleixner et al. 1998; Gessler et al. 2008) and stems

(Skomarkova et al. 2006). Storage effects are most

important at the end and at the beginning of the growing

season (Helle and Schleser 2004; Scartazza et al. 2004;

Schulze et al. 2004), although the interactions between

storage products and growth are not fully understood.

A carry-over effect may lead to autocorrelations among

years and mask a climatic signal (Hill et al. 1995;

Monserud and Marshall 2001; Kagawa et al. 2006a, b).

The aims of the study reported here were (1) to determine

the correlation between climate parameters and tree-ring

width (TRW) and wood anatomy (wood density, cell-wall

thickness, cell-lumen area) at three sites ranging between

boreal and sub-Mediterranean alpine climates; (2) to com-

pare the long-term (decadal) correlations between tree rings

and climate parameters with short-term observations; (3) to

analyze the seasonal variation of tree-ring d13C in trees

exhibiting different growth rates; (4) to investigate relations

between present and past years on tree-ring parameters.

Materials and methods

Experimental sites and tree samples

The three study sites are (Table 1; see also http://www.

bgc-jena.mpg.de/public/carboeur/sites/index-s.html): (1) an

alpine forest in North Italy (Renon, REN), (2) a lowland

forest in Germany (Hainich, HAI) and (3) a boreal forest in

North Sweden (Flakaliden, FLA). Annual temperatures

were highest in HAI and lowest in FLA, and the length of

the growing season ([5�C) was longest (211 days) in

Germany, 20% shorter at the alpine site and 34% shorter at

the boreal site. Annual precipitation and summer rainfall

decreased from the alpine REN site to the boreal FLA site.

The investigated stands differed in age and stand density.

The distribution of forests at the global scale is related to

precipitation, but the growth of trees is not only determined

by climate but also by nutrition (Linder 1995; Schulze et al.

2006). Table 1 shows that the boreal site has a much lower

nitrogen concentration in its needles than the other sites

and that the needle nitrogen level reached at FLA indicates

deficiency (Oren and Schulze 1989).

Soil conditions differed between sites. The boreal site at

FLA is a glacial moraine with a shallow organic layer on

unsorted stony material in the whole soil horizon. The

lowland site at HAI is a brown earth developed on clay that

originated from the weathering of limestone. The alpine

site REN is a shallow organic layer that developed on

limestone boulders.

In accordance with the climate analysis of Walter and

Lieth (1967), all sites have a positive hydrological balance.

During the time interval studied in this investigation

(Table 2; 1999–2003), the seasonal climate showed very

little variation in early season temperatures, but tempera-

tures were colder during the autumn at the boreal site.

Average daytime vapor pressure deficit was highest at the

HAI site. Early season precipitation was lowest at the

boreal site, but low rainfall in the spring is generally

considered to be compensated for by snow melt in boreal

climates. Late season rainfall was lowest at HAI. Solar

radiation was highest at the alpine site REN. Due to the

long daylight in the northern summer, solar radiation was

higher at FLA than at HAI in the early growing season, but

this decreased in the late summer.

The seasonal averages do not show the inter-annual var-

iability of climate, which is important to this study (Fig. 1).

Maximum temperatures increased at all sites between 1999

and 2003. The length of the growing season (days with

average temperature[5�C) did not show large inter-annual

variability, while precipitation showed the largest variabil-

ity, with drought being more apparent later in the season.

Years with precipitation during the growing season below

the long-term average were considered to be dry years. At

FLA, 1999 and 2002 were considered to be dry years, with

rainfall up to 30% below average; at HAI and REN, 2001

and 2003 were considered to be meteorologically dry years.

Since the CarboEurope climate data cover only 5 years,

additionally long-term meteorological data (daily temper-

ature and precipitation) were used from the following

730 Oecologia (2009) 161:729–745

123

weather stations: Costalovara (46�520N, 11�420E, 1,250 m

a.s.l., data for 1989–2003), Soprabolzano (46�520N,

11�400E, 1,206 m a.s.l., data for 1931–1980), Leinefelde

(51�200N, 10�220E, 440 m a.s.l., data for the 1957–2003)

and Stensele (65�040N, 17�100E, 330 m a.s.l., data for

1918–2001).

Tree sampling and dendrochronological analysis

Wood samples were collected by coring one core per tree at

breast height (1.3 m) with a 5-mm-diameter increment

borer. At Renon, eight cores of Norway spruce [Picea

abies (L.) H. Karst] and three cores of Swiss stone pine

(Pinus cembra L.) were used to measure wood density, and

five cores of each species were used to measure d13C and

anatomy. For Hainich, three cores each of Norway spruce

were used to measure d13C and anatomy, respectively. For

Flakaliden, three cores each of Norway spruce and Scotch

pine (Pinus sylvestris L.) were used for d13C and anatomy,

respectively. The statistical data of wood density and

tree-ring width chronologies are based on International

Tree-Ring Data Bank (ITRDB; FH Schweingruber) for the

sites in Germany and Sweden.

Tree-ring width was measured on all cores using a semi-

automatic device LINTAB-III (Cook and Kairiuktis 1990;

Rinn 1996; Vaganov et al. 1996). Density profiles of tree

rings were measured following Schweingruber (1988)

using a densitometer (Dendro-2003; Walsch Electronics,

Zurich, Switzerland).

We also used the existing in ITRDB tree-ring width and

maximum density chronologies obtained by FH Schwe-

ingruber at the end of the 1970s for sites close to the study

areas. The analog sites of HAI and FLA are Andreasberg

Harz Schlucht (900 m NN) and Gallejour Glommerstark

(480 m NN), respectively (http://www.ncdc.noaa.gov/

paleo/ftp-treering.html).

Cross-dating of individual cores was carried out based

on tree-ring width (Schweingruber 1988) using the pro-

gram COFECHA (Cook and Peters 1981; Holmes 1992).

Absolute values of each tree-ring width were standardized

by calculating the ratio of the tree-ring deviation in a

certain year and the long-term average of tree-ring width to

remove the effect of age and diameter on tree-ring width

(Shiyatov 1986; Cook et al. 1990; Vaganov et al. 1996).

A spline function was used to determine the long-term

trend for each tree core using the program ARSTAN (Cook

et al. 1990). The following parameters were calculated

according to Schweingruber (1988) as:

Table 1 Climate and stand characteristics of the Renon site (Cescatti and Marcolla 2004), the Hainich site (Knohl et al. 2003) and the Flakaliden

site (Linder 1995; Lundmark et al. 1998)

Site Renon (North Italy) Hainich (Central Germany) Flakaliden (North Sweden)

Longitude 46�360 51�040 64�070

Latitude 11�280 10�270 19�270

Elevation (m) 1,730 445 310

Length of growing seasona ([5�C) (day) 30 April–14 October (167) 31 March–27 October (211) 9 May–27 September (141)

Average annual temperaturea (�C) 3.8 6.8 1.9

Precipitation, annual (mm) 1,008 780 590

Species (no. of samples) Picea abies (13) Pinus cembra (8) P. abies (5) P. abies (5) Pinus sylvestris (5)

Age (years) 182 80 31

Diameter at breast height (cm) 17 40 8

Height (m) 29 31 6

Stand density (trees/ha) 745 679 2100

Needle nitrogen concentration (mmol g-1) 0.84 0.94 0.68

a Temperature data were corrected according the differences between the elevation of the studied site (1700 m) and that of the meteorological

station (1210 m)

Table 2 Comparison of early and late season climate between sites

Parameter Renon Hainich Flakaliden

Temperature–MJJ (�C) 14.8 15.2 11.6

Temperature–ASO (�C) 12.5 14.8 7.8

Vapor pressure deficit–MJJ (hPa) 4.2 5.8 4.7

Vapor pressure deficit– ASO (hPa) 3.7 4.5 2.6

Precipitation–MJJ (mm) 324 214 209

Precipitation–ASO (mm) 242 177 202

Precipitation-growing season (mm) 566 391 411

Solar radiation-MJJ (MJ m-2) 967 621 665

Solar radiation–ASO (MJ m-2) 686 491 299

MJJ, May–June–July; ASO, August–September–October

Study interval: 1999–2003

Oecologia (2009) 161:729–745 731

123

1. The standard deviation of relative TRW and maximum

density, indicating a measure of tree-ring response to

inter-annual climate variability.

2. The ‘‘interserial correlation’’ describes the variation in

a tree-ring parameter in a certain year between all

sample trees of a stand. This parameter provides

information on the heterogeneity of the stand and

indicates the similarity of tree growth during inter-

annual changes in climate. A high number indicates a

high similarity between trees.

3. The common variance of the first eigenvector

explains the fraction of the variability in tree-ring

parameters by environmental factors common to all

trees at each site (Fritts 1976; Cook et al. 1990).

A high number indicates a strong climatic effect on

all trees of a site.

4. The first-order autocorrelation explains the correlation

between the tree-ring width in the previous year (t-1)

and ring width in the year under investigation (t).

A large number represents a strong effect of the past

year on the current year’s tree ring.

The effect of temperature and precipitation on the inter-

annual variability of tree-ring growth and wood density

was estimated through correlations between the chronolo-

gies and monthly climatic data from the above mentioned

meteorological stations. These calculations were performed

for each month of 12 months (from October of the previous

year to September of the current year) using the program

RESPONSE (Holmes 1983).

Image analysis of tracheid dimensions

Wood cross sections were cut with a microtome (20 lm)

and stained by methylene blue (Furst 1979; Vaganov et al.

1985). Tracheid dimension in tree rings was determined

using an Image analysis system (Carl Zeiss, Jena; Munro

et al. 1996; Vaganov et al. 2006). Radial cell diameter (DR),

cell-wall thickness (CWT) and tangential cell diameter

(DT) were measured in five radial series of cells from the

outside border in each tree ring to the inside border for each

year. The five radial series were 100–150 lm (three to five

cell rows) apart. Assuming a rectangular shape of the

tracheid in cross section, these data were used to calculate

cell-wall area: CWTarea ¼ 2CWT ðDT þ DR � 2CWTÞ,and cell-lumen area: LUMarea ¼ DR � DT � CWTarea.

Since tree-ring width varies between years and species,

measurements were normalized to a standard number of

cells (Vaganov 1990).

Seasonal course of anatomy and isotopes

Tracheids of conifers are organized in rows in which each

tracheid was produced by the cambial zone consecutively

after the previous one. As such, the position (from the first

early wood cell to last late wood cell) of each tracheid in a

row is related to a temporal scale. In order to fit the ana-

tomical data measured within a tree ring with calendar

dates (dates of season) for some samples, we used the

approach described in detail by Vaganov et al. (2006)

1999 2000 2001 2002 2003

Hainich

A M J J A S O0

50

100

150

200

A M J J A S O A M J J A S O A M J J A S O A M J J A S O0

5

10

15

20470 mm 343 mm 442 mm 594 mm 236 mm

Renon

A M J J A S O

Pre

cipi

tatio

n (m

m)

0

50

100

150

200

A M J J A S O A M J J A S O A M J J A S O A M J J A S O

Air

Tem

pera

ture

(°C

)

0

5

10

15

20753 mm 661 mm 563 mm 692 mm 470 mm

Flakaliden

A M J J A S O0

50

100

150

200

A M J J A S O A M J J A S O A M J J A S O A M J J A S O0

5

10

15

20356 mm 565 mm 557 mm 290 mm 422 mm

Fig. 1 Inter-annual variability in temperature and rainfall at the three study sites for the period of 1999–2003. Gray shadow indicates dry years,

horizontal bars indicate the meteorological length of the growing season ([5�C daily temperature)

732 Oecologia (2009) 161:729–745

123

which transforms the tracheid radial size to the dates of

their production by the cambium based on two main

assumptions: (1) a linear relationship between growth rate

and radial tracheid size, and (2) a known duration of the

growth season. In tree-ring studies, the length of the growth

season is the period with an average daily air temperature

above 5�C (Mikola 1962; Leikola 1969; Hughes et al.

1999; Jones and Briffa 1992; Kirdyanov et al. 2003). Based

on this assumption, the linear sequence of tracheid mea-

surements is transformed into a seasonal time scale for

each consequent tracheid in the row, accounting for a delay

of 25 days, which is needed for cells to reach a final

diameter and to start with secondary cell-wall growth

(Vaganov et al. 2006).

Carbon isotope analysis

Carbon isotope ratios were determined using a laser abla-

tion-combustion line coupled to an isotope-ratio-mass-

spectrometer (YAG 266 nm UV laser; Merchantek–New

Wave, Fremont, CA, coupled to a Finnigan Delta ?XL) as

described in detail by Schulze et al. (2004). The exact

location of each ablation spot was visualized with a camera

mounted directly on the laser ablation station. The wood

samples were enclosed in an aluminum chamber with a

quartz window, which was flushed with helium as carrier

gas. For quantitative combustion of the ablation particles to

CO2 and H2O, the sample was passed through an Al2O3

tube at 700�C containing CuO wire as the oxygen source.

The reaction gas further passed through a gas chromatog-

raphy (GC) column (HayeSep D, Bandera, TX) to separate

CO2 from other gases. Water was removed with an on-line

Nafion water trap. Carbon-stable isotope ratios were

expressed in the d notation on the VPDB scale using

NBS22 with a value of -30.03% as the scale anchor. The

spatial resolution of a laser shot is about 70 lm. A series of

four to five shots was needed at one location to collect

sufficient material for analysis. The series of shots was

repeated radially along the same line every 120 lm, which

resulted in ten to fifteen data points per tree ring. Profiles of

d13C were measured for a 19-year period.

The d13C values within a tree ring varying within and

between years were normalized by calculating the devia-

tion of the measured value at time t from the average of

that same year (d13Ct - d13Cavt).

In a separate experiment, the d13C value of ray cells

and of adjacent tracheids were measured in the last cells of

late wood and in the first cells of early wood in order to

detect changes which may be related to carbohydrate

storage.

We use the d13C notation rather than discrimination

(D) because we did not measure the atmospheric d13C.

Furthermore, the reconstruction of all discrimination steps

during wood formation was not the objective of our study

and, therefore, the comparison of d13C values seemed to be

more appropriate. The d13C-values represent the isotope

ratios of natural wood, including lignin. Schulze et al.

(2004) showed that there is an offset between cellulose and

natural wood of 1.5% in conifers, which is constant and

independent of season. Eglin et al. (2008) also confirmed

that biochemical composition influences the variability in

the carbon isotope composition of tree rings by up to a

maximum of only 5%. Consequently the investigation of

whole wood samples seems to be appropriate (Harlow et al.

2006; Eglin et al. 2008).

Statistical analysis

Relationships between wood anatomy and carbon isotope

ratios and their dependency on climatic variables were

analyzed with linear mixed-effects models using the lmer

function in the R package lme4 (Bates and Sakar 2006) to

account for unbalances in the data set due to different

numbers of investigated trees per site and study years.

A hierarchical series of models was fitted sequentially,

entering site, wood anatomical or climatic variables and

their interactions, with site as fixed effects. Tree identity

and study year were included as grouping factors in a

nested sequence in the random term. The maximum like-

lihood method was applied, and likelihood ratio statistics

were used to appraise model improvement and to test for

statistical significance of the fixed effects.

Results

A typical pattern of the anatomical parameters and carbon

isotope ratios of the 2001 tree ring of spruce from REN

(Fig. 2) shows that the radial size of cells decreases slowly

in early spring wood and faster in late early wood and late

wood. Early and late wood are clearly visible in the

density slide. At the same time, CWT and area increase,

reaching a maximum in the first cells of late wood and

decreasing again before the cessation of growth at the end

of the season. Changes in lumen area parallel the changes

in radial cell size. Wood density can be seen to increase in

early wood slower than in late wood, reaching a maximum

in the last cells of late wood. Values of d13C increase

gently in early wood from the beginning of the growth

ring to the transition zone between early and late wood

and then decrease sharply in late wood. This decrease in

d13C (more negative values indicate open stomata) in late

wood was unexpected because precipitation was low in

August.

Oecologia (2009) 161:729–745 733

123

Long-term relations between tree-ring width, maximum

density and climate

Standardized tree-ring width and maximum wood density

of spruce and pine species varied with a standard deviation

of up to 20%, and they were similar between both species

studied (Table 3; Fig. 3). Both species showed higher

inter-annual variability for tree-ring width than for maxi-

mum wood density. The standard deviation of the

standardized tree-ring width, indicating inter-annual vari-

ability, was lower at the alpine REN site [standard devia-

tion (SD) 10%] than at the boreal FLA and the lowland

HAI site (SD 20%; Table 3). Inter-annual variability of

wood density was very small (6–7%) and similar for all

sites and species.

The inter-serial correlation (Table 3) that compares

the variation between trees is similar for maximum density

and for tree-ring width. The standard deviation of the

0

100

200

300

400

500

600

0

200

400

600

800

1000

1200

Picea abies (sample 14)

δ13C

(‰)

-23.5

-23.2

-22.9

-22.6

-22.3

Rad

ialc

ells

ize

(mm

)

0

10

20

30

40

50

Cel

lwal

l thi

ckne

ss(µ

m)

0

1

2

3

4

5

6

7

Date

Cel

lwal

lare

a(µ

m2 )

Lum

enar

ea(µ

m2 )

Woo

dde

nsity

(mg

cm-3)

Radial cell size

Cell wall thickness

Lumen area

Wood density

Cell wall area

May Jun Jul Aug Sep2001

Fig. 2 Example of anatomical,

density and isotopic

measurements in a spruce tree

ring (Renon, Italy; Picea abies,

tree number 14): radial cell size,

cell-wall thickness, lumen area,

cell-wall area, wood density.

The laser shots for isotope

determination can be seen as

vertical bars

734 Oecologia (2009) 161:729–745

123

Table 3 Statistical parameters of tree-ring width and maximum density chronologies for conifers from the studied sites

Species Parameter Site Number

of samples

Standard deviation of

relative TRW and MaxD

Inter-serial correlation

(average ± SD)

Common variance

(first Eigen-vector)

First-order

autocorrelation

Picea abies TRW REN 13 0.10 0.32 ± 0.20 32.3 0.75

TRW HAIa 14 0.20 0.68 ± 0.21 61.2 0.54

TRW FLAa 13 0.19 0.71 ± 0.46 61.1 0.76

MaxD REN 8 0.06 0.60 ± 0.06 54.2 -0.16

MaxD HAI 14 0.07 0.74 ± 0.08 69.7 0.03

MaxD FlA 13 0.07 0.64 ± 0.09 – –

Pinus cembra TRW REN 8 0.16 0.28 ± 0.18 35.2 0.72

Pinus sylvestris TRW FLA 18 0.17 0.58 ± 0.23 55.3 0.58

Pinus cembra MaxD REN 3b 0.06 0.30 ± 0.06 37.7 0.19

Pinus sylvestris MaxD FLA 18 0.07 0.67 ± 0.09 55.2 0.21

TRW, Tree-ring width; MaxD, maximum diameter; SD, standard deviation; REN, Renon (Italy), an alpine forest in North Italy; HAI, Hainich, a

lowland forest in Germany; FLA, a boreal forest in Northern Sweden (Flakaliden, FLA)a Data from ITRDB (FH Schweingruber)b Number of samples is not enough for significance

Maximum wood density

1900 1920 1940 1960 1980 20000.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

1750 1800 1850 1900 1950

Sta

ndar

dize

d m

axim

um d

ensi

ty

Years1900 1920 1940 1960 1980

Tree ring width

1900 1920 1940 1960 1980 2000

Renon (Italy)

1750 1800 1850 1900 1950

Sta

ndar

dize

d tr

ee r

ing

wid

th

Hainich (Germany)

Years1900 1920 1940 1960

Flakaliden (Sweden)

Pinus cembra

Picea abiesPinus cembra

Picea abies

Picea abies Picea abies

Picea abies

Picea abies

Pinus sylvestris

Pinus sylvestris

Fig. 3 Standardized site tree-

ring width and maximum

density chronologies for Piceaabies and Pinus cembra. The

original data were standardized

by correcting for an age-

dependent trend

Oecologia (2009) 161:729–745 735

123

inter-serial correlation was higher for tree-ring width than

for maximum wood density, and it was highest at FLA and

lowest at REN. Thus, there was relatively little variation

between trees at FLA and HAI, but a large variation

between trees at REN. Also, the common variance of the

maximum density and tree-ring width was higher at FLA

and HAI than at REN, indicating a weaker climatic signal

at REN.

The first-order autocorrelation in tree-ring width, which

is an indicator of the carry-over effect from one year to the

next, is very high at all study sites. In fact, 29–58% of tree

ring-width in a current year is explained by the growth rate

in the previous year (Table 3). The autocorrelation in the

tree rings is much larger than the autocorrelation of cli-

matic parameters, which was 0.06, 0.08 and 0.16 for tem-

perature, and 0.10, -0.12 and -0.06 for precipitation at the

REN, HAI and FLA site, respectively (data not shown).

This result indicates that only 0.3–2% of the climatic

conditions in one year are explained by the climatic con-

ditions in the previous year.

The correlation of long-term chronologies of tree growth

and wood density with temperature and precipitation (time

series 70–100 years) shows no general pattern between tree-

ring width or maximum wood density and monthly mean

values for precipitation and temperature (Fig. 4). Site-

specific conditions are important. For example, late season

temperature (August, September, October) correlated with

tree-ring width at the alpine REN and boreal FLA sites, but

the early season temperature (May, June, July) was signifi-

cant at the lowland HAI site. Precipitation had no significant

effect on tree-ring width at the alpine site, but late season

precipitation correlated with tree-ring width at HAI, and

early season precipitation was significant at FLA despite

snow melt. In general, there is a stronger climatic effect at

higher latitudes (Shiyatov 1986; Vaganov et al. 1996; Briffa

et al. 1998) than at the alpine site. The climate signal was

not as strongly expressed as that observed for maritime

Mediterranean sites (Gonzales and Eckstein 2003).

In summary, we found that the long-term correlations

between climate and tree-ring width and wood density are

not very strong and that they are both site- and species-

specific. For tree-ring width, late season temperatures (but

not precipitation) are significant at REN, and spring tem-

peratures and precipitations are significant at HAI and FLA.

Extreme events of temperature and precipitation, such as in

the dry year of 2003 (see Fig. 7), seem to be important.

Relations between isotopic composition and cell lumen

and tree-ring width

All sites show a positive correlation between the average

value of d13C within a tree ring and the width of tree rings

(Fig. 5). This was unexpected because a more positive

d13C value would indicate an increasing stomatal limitation

of photosynthesis with increasing ring width. The main

differences between sites are the intercept values and a

lower slope at the REN site. In narrow rings, the temperate

and the boreal site are more similar than the alpine and the

boreal site. The high d13C values at REN could indicate

water stress at the alpine site. This was not apparent in the

correlations with long-term chronologies (Fig. 4).

The variations in d13C and wood anatomy within a tree

ring were investigated for the REN site (Fig. 6) by calcu-

lating the deviation of d13C from the average value for a

given tree ring. Cell lumen area was used as a parameter

because it is positively correlated with tracheid diameter

(overall correlation 0.86, P \ 0.0001), and it is negatively

correlated with CWT (overall correlation -0.48,

P \ 0.01), cell-wall area and density (Lambers et al. 1998;

Hubbard et al. 2001).

Cell-lumen area shows the highest values at the begin-

ning of the growing season and continually decreases later

in the season. The decrease is steeper in narrow than in

wide rings. The maximum values of cell lumen area are

slightly higher for spruce (1063–1225 lm2) than for pine

(956–1058 lm2). There is no obvious change at the border

of early and late wood.

The seasonal changes of d13C with tree-ring width can

be seen to differ from the seasonal changes in cell-lumen

area (Fig. 6). At the beginning of the season, average d13C

was 0.4–0.8% lower than the tree-ring average; it then

increased with progressing tree-ring formation (see Fig. 5),

but decreased again past a maximum. This decrease may

only be small in narrow rings, but it is substantial in wide

tree rings. Thus, in wide tree rings, the late wood d13C may

be lower than in the early wood, while in narrow rings d13C

in late wood is higher than in early wood of the same ring.

The average amplitude of the intra-annual variation of d13C

is about 0.8–1.6% depending on years of growth.

The changes in d13C within the tree rings (see Fig. 6) do

not correlate with the observed long-term effect of climate

on tree-ring width (Fig. 4) and may reflect seasonal chan-

ges in the contribution of storage products for tree-ring

growth.

The seasonal and inter-annual variation

of d13C in spruce

In this section, we depict examples of the seasonal changes

in d13C in fast-, average- and slow-growing trees of Picea

abies with the aim of demonstrating the tree-by-tree vari-

ability as well as inter-annual variability for the three sites

(Fig. 7). The different numbers of years in Fig. 7 are due to

the sampling procedure, where cores of different lengths

were taken from fast- and slow-growing trees. Thus, not all

cores reach the same depth from the cambium.

736 Oecologia (2009) 161:729–745

123

Figure 7 shows that the variation in d13C within years

(intra-annual variation) is highly variable between fast- and

slow-growing trees. d13C was either constant, increased or

decreased over time, with no consistent relations.

In fast-growing trees, the d13C of early wood was on

average the same for all three sites (24.9 ± 0.6%). The

seasonal course of d13C in fast-growing trees at REN

showed an autumn decrease, even in the dry years of 2001

and 2003. At HAI, dry years (2001 and 2003) differ from

wet years in that there was an increase in autumn d13C in

the dry years. At FLA also, there may be an increase in

d13C in the autumn of wet years, and there was no d13C

signal left in fast-growing trees of FLA during the very dry

year of 2002.

Slow-growing trees d13C showed a more consistent

pattern than fast-growing trees. At REN, d13C generally

increased as the season progressed, which was opposite to

the seasonal pattern observed in fast-growing trees.

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

Cor

rela

tion

coef

ficie

nts

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

Tree ring width

Cor

rela

tion

coef

ficie

nts

Maximum wood density

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

Hainich (Germany)

Flakaliden (Sweden)

Picea abies Picea abies

Picea abies Picea abies

Months

Renon (Italy)

MonthsOct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep

Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep

Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep

Pinus cembra Pinus cembra

* *

Picea abies Picea abies

* *

1950-1998

**

*

**

**

*

*

**

**

*

*

* * * * *

1957-2003

1901-1998

T T

T

T

T

T

T

TP

P

P

P

PP

P

P

Fig. 4 Correlation coefficients

of annual tree-ring and

maximum density chronologies

with climatic data: Linesindicate correlations with

monthly mean temperature,

columns indicate correlations

with monthly precipitation.

Asterisks indicate a significant

correlation (P \ 0.01)

Oecologia (2009) 161:729–745 737

123

At HAI, the effect of a dry year was not apparent in slow-

growing trees, while at FLA, the very dry year of 2002 did

not leave a signal in slow-growing trees, but the lesser dry

year of 1999 and 1997 resulted in a increase in d13C.

The highest values of d13C were found at REN for a tree

of average growth rate (-22.8%) with very little inter-

annual and seasonal variation. The d13C values were gen-

erally lower in slow-growing trees than in fast-growing

ones. The lowest values were observed in slow-growing

trees of HAI and FLA (-27%). It should be pointed out

that all trees were dominant in height and that none of the

trees was a suppressed tree of the lower canopy.

There was a remarkable stepwise change in some trees

in terms of d13C between late wood and early wood of the

next year. A decrease in d13C across the tree ring boarder

was observed at REN, where the lowest d13C values were

observed in autumn. This result contrasts with those

obtained at the HAI and FLA sites, where the d13C values

of the early wood were generally lower than those of the

late wood in autumn.

In order to obtain insight into the processes which link

the late wood of one year with the early wood of the next

year, we assessed whether the d13C in ray cells differed

from the d13C in xylem cells (Table 4). The expectation

was that ray cells, which contain the stored carbohydrates

in conifers, would contain an additional isotope signal; i.e.

the hypothesis was that ray cells would be heavier that the

associated tracheids. However, at all sites, the d13C in 1- to

3-year-old ray cells was 0.1–0.2% lower than that of the

associated xylem. This difference disappeared after

4 years, i.e. ray cells were the same as xylem cells. A lower

d13C value could be the result of a continuation of lignin

incorporation into ray cells (FH Schweingruber, personal

communication), which would lower the d13C of ray cells.

Obviously, the bulk measurement of the whole cell is not

fully conclusive and does not show the effects of carbo-

hydrate turnover.

Independent of sites, yearly climate and growth rates

(Fig. 8), we found a close correlation between the yearly

average d13C of a tree ring in a given year (t) and the

Tree ring width (mm)0.0 0.5 1.0 1.5 2.0 2.5 3.0

δ13C

(‰

)

-28

-27

-26

-25

-24

-23

-22

Flakaliden

Hainich

Renon

Renon: y=-24.33+0.39x, r2=0.053, p=0.165

Hainich: y=-26.16+0.73x, r2=0.221, p=0.024

Flakaliden: y=-26.97+0.81x, r2=0.375, p=0.003

Fig. 5 Correlation between tree-ring width and average value of d13C

within tree rings for small (\3 mm) tree rings. The average d13C

value represents the numerical average of the individual measure-

ments without taking changes in density into account. The higher

density of late wood affects only 10–20% of the d13C data

Picea abies

0 1 2 3 4 0 1 2 3 4

0 1 2 3 4 0 1 2 3 4

0

200

400

600

800

1000

1200

1400narrow; y=916.90-0.128x-0.0011x2, r2=0.785

middle; y=1027.90-0.304x-0.0002x2, r2=0.745

wide; y=943.14-0.032x-0.0002x2, r2=0.788

very wide; y=890.07-0.043x-3.99e-5x2, r2=0.890

Pinus cembra

Cel

l lum

en a

rea

(µm

2 )

0

200

400

600

800

1000

1200

1400narrow; y=872.89-0.705x-7.59e-5x2, r2=0.620

middle; y=893.17-0.176x-0.0003x2, r2=0.777

wide; y=947.58-0.164x-9.58e-5x2, r2=0.865

Distance from tree ring border (mm)

narrow; y=-0.456+0.0015x-8.78e-7x2, r2=0.327

middle; y=-0.461+0.0013x-7.04e-7x2, r2=0.374

wide; y=-0.207+0.0008x-4.04e-7x2, r2=0.374

very wide; y=-0.125+0.0006x-2.05e-7x2, r2=0.654

δ13C

-δ13

Cav

erag

e (‰

)

Cel

l lum

en a

rea

(µm

2 )δ13

C-δ

13C

aver

age

(‰)

-2.0

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

-2.0

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

narrow; y=-0.712+0.0032x-2.50e-6x2, r2=0.596

middle; y=-0.680+0.002x-1.08e-6x2, r2=0.484

wide; y=-0.849+0.0019x-7.62e-7x2, r2=0.533

Renon RenonFig. 6 Relations between cell

lumen area and d13C with tree-

ring width as expressed by

distance from the tree ring

boundary. The Renon site was

taken as an example

738 Oecologia (2009) 161:729–745

123

average d13C in the immediately preceding year (t-1). At

low d13C values, there was a tendency for the yearly

average d13C of early wood to be higher than that of late

wood; in contrast, at high d13C values, the yearly average

d13C in early wood would tend to be lower than that in late

wood.

It was not possible to compare d13C in wood with eddy

flux data at all sites because the ecosystem flux at the

alpine site REN is also determined by the activity of the

grass cover on the forest floor, while the stand was too

small at the lowland HAI site for eddy flux measurements.

At the boreal site FLA, it was only possible to make a

comparison for 2001 and 2002 before the eddy flux mea-

surements terminated. The synchronization with growth

was made on the basis of independent dendrometer mea-

surements (Sune Linder, personal communication). We

used gross primary productivity and canopy conductance to

derive a canopy-scale ratio of the intercellular to ambient

CO2 concentration (ci/ca), which is not equal to leaf–level

ci/ca, but rather a integrative estimate of ci/ca across the

entire canopy confounded by other effects, such as surface

evaporation and mixing sun and shade leaves. Figure 9

shows an increase in d13C early in the season in both years,

when canopy-scale ci/ca was increasing. Thus, the initial

increase in d13C is in contrast to the expectation that gas

exchange parameters determine the isotope composition in

wood. Also, the d13C in the following year was lower or

similar to that in the past autumn, which was contrasted by

much lower ci/ca values. Thus, gross primary productivity,

canopy conductance and canopy-scale ci/ca do not explain

the observed variation in d13C in wood.

The information obtained during our study, which is

based on a total of three sites, 22 years of wood analysis

and 760 individual observations, was analyzed with linear

mixed-effects models to explore relationships between

d13C, wood anatomy, yearly climate and their dependency

on different study sites (Table 5). Cell-wall area was

related to cell lumen in early and late wood, and d13C

correlated with cell-wall area and lumen in early and late

wood even if the significant interaction terms with site

identity indicated differences in these relationships among

sites. The relationships to climate parameters are compli-

cated and mostly site-specific. Tree-ring width correlated

with early season precipitation and late season temperature

(mainly an effect of HAI). Vapor pressure deficit (VPD)

and the total incoming radiation (R) were highly significant

variables to explain variation in tree-ring width although

their effects varied to some degree among sites as well.

d13C varied among sites, but it generally did not correlate

with precipitation or temperature. However, despite site-

specific differences, d13C was highly significantly related

to total radiation, VPD and the length of the growing

HAI-Sp3015

0

HAI-Sp3014

REN-Sp8

024681012 24681012 24681012

2000 2001 2002 2003

REN-Sp4

Distance from cambium (mm)

2002

2003

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

1998 1999 2001 2002 20032000REN-Sp14 FLA-Sp9

1999 2000 2001 2002 2003

FLA-Sp3

δ13C

(‰

)

1998

1999

2000

1997

2001

2002

FLA-Sp1

0

0024681012 24681012 24681012 0

0024681012 24681012 24681012 0

-28

-27

-26

-25

-24

-23

-22

-28

-27

-26

-25

-24

-23

-22

-28

-27

-26

-25

-24

-23

-22

20032002200120001999

2000 2001 2002

2003

2004

HAI-Sp30162000 2001 2002 2003 2004

2004200320022001

earlywoodlatewood

tree

with

rap

id g

row

thav

erag

e gr

owth

slow

gro

wth

Fig. 7 Seasonal courses of d13C during consecutive years, taking a

fast-growing tree, a tree with average growth rate and a slow-growing

tree as an example. Closed symbols indicate early wood, open

symbols indicate late wood. Columns: left Renon (REN), middleHainich (HAI), right Flakaliden (FLA)

Oecologia (2009) 161:729–745 739

123

season. The correlation between early wood of one year

versus late wood of the immediately preceding year was

highly significant.

Figure 10 summarizes the yearly averages of all data

points in terms of yearly d13C and tree-ring width. There is

a strong latitudinal effect of d13C, which was highest at

REN and lowest at FLA. Although REN receives 20–40%

more solar radiation than HAI and FLA, respectively, tree-

ring width at REN varies mostly between that at HAI and

FLA. The HAI site receives only 30% more solar radiation

than FLA, but the tree-ring width was found to be twofold

larger. Despite high rainfall, REN appears to be a water-

stressed site due to the special soil conditions, but water

stress alone cannot explain the differences. Tree-ring width

is lowest in Flakaliden, despite low d13C. There are obvi-

ously additional factors which determine tree-ring width

and d13C, one of which is plant nutrition. FLA is nitrogen

deficient (Table 1), while the nitrogen concentrations of

needles were the highest at REN. High nitrogen availability

is expected to increase carboxylation capacity and to

amplify water stress, leading to an increase of d13C in leaf

and wood (Hogberg et al. 1993; Livingston et al. 1999) and

thus explain part of our observation.

Discussion

This study was undertaken to investigate correlations

between climate and tree-ring width and anatomy at long

and short time scales and to explore whether d13C could be

used as an additional parameter to interpret tree-ring chro-

nologies. We also wanted to know if there are carry-over

effects from one year to the next that affect d13C and

growth. We investigated these questions by (1) studying

climate correlations of long-term tree ring chronologies, (2)

investigating in detail short-term seasonal observations

using wood samples which were taken from forest sites

ranging between boreal Scandinavia and alpine North Italy

where trees grew in competitive interaction as forest stands.

In the long time series, temperature and precipitation

correlated with tree-ring width and wood density, but these

correlations were highly site and species specific, with

spring-time temperatures and precipitations seeming to be

most important. In contrast to the long-term correlations, at

short time scales, we found only poor correlations between

tree-ring width and temperature or precipitation. However,

there was a highly significant correlation between tree-ring

width, d13C and lumen area and the length of the growing

season, total net radiation and VPD. Our investigation also

shows that there were large differences between fast- and

slow-growing trees even though all trees were dominant

canopy trees in which the d13C signal was not affected by

soil respiration. It is quite likely that the main driver for

d13C could be VPD. We were unable to test for effects of

soil moisture because these data were not available as time

series for all sites.

Our initial hypothesis was that the seasonal course of

d13C would allow an identification of dry years (i.e. years

Table 4 Difference between d13C of xylem and ray cells

Year Early (EW)-Late

wood (LW)

Hainich

d13C-xylem

Hainich

d13C-ray

Hainich d13C

xylem-ray

Renon

d13Cxylem

Renon

d13C-ray

Renon d13C

xylem-ray

Flakaliden d13C

xylem-ray

2001 EW -25.51 ? 0.42 -25.98 ? 0.42 -0.01 ? 0.02

LW -25.44 ? 0.79 -25.54 ? 0.80 0.11 ? 0.14

2002 EW -25.62 ? 0.56 -25.69 ? 0.58 0.07 ? 0.17 0.16

LW -25.84 ? 0.37 -25.95 ? 0.31 0.11 ? 0.12 -24.66 ? 1.20 -24.68 ? 1.20 0.01 ± 0.05 0.21

2003 EW -25.37 ? 0.62 -25.60 ? 0.68 0.23 ? 0.13 -24.36 ? 0.85 -24.48 ? 0.83 0.12 ± 0.13

LW -23.91 ? 1.39 -24.10 ? 1.34 0.18 ? 0.12 -24.31 ? 1.35 -24.40 ? 1.35 0.08 ? 0.15

2004 EW -26.11 ? 0.61 -26.26 ? 0.69 0.15 ? 0.12

LW -25.44 ? 0.79 -25.54 ? 0.80 0.09 ? 0.07

δ13C

ew (

curr

ent y

ear) (

‰)

δ13Clw (previous year)

(‰)

Picea abies

-28 -27 -26 -25 -24 -23 -22-28

-27

-26

-25

-24

-23

-22

RenonHainichFlakaliden

y=-5.83+0.78x, r2=0.626, p=0.0000

1:1

Fig. 8 Relation between d13C in late wood of previous tree ring and

in early wood of current tree ring for spruce. The thin line shows the

1:1 relation. The regression lines are calculated by reduced axis

regression

740 Oecologia (2009) 161:729–745

123

with below-average rainfall during the growing season) as

had been described by Ferrio et al. (2003) in Pinus

halepensis, Kagawa et al. (2003) in Larix gmelinii or

Adams and Kolb (2004) in Pinus ponderosa. We observed

a d13C response in a fast-growing tree of HAI during the

dry year of 2003; however, the effect disappeared in slow-

growing dominant trees, and the d13C increase did not

increase at the other sites during years with a below-

average rainfall. To the contrary, d13C increased even in

wet years at FLA. d13C did not respond at REN. There is

clearly an absence of a general and regular response of

tree-ring width and d13C to precipitation. The site effects

and the effects of slow- versus fast-growing trees override

the precipitation effect.

The seasonal increase in d13C at FLA was not caused by

drought. Rainfall at FLA is generally lower in August, but

this did not affect canopy-scale ci/ca, and the d13C of wood

changed independently of canopy-scale ci/ca. We hypoth-

esize that the increase in d13C in autumn is caused by the

reserve formation of products which are 13C-enriched and

which then result in heavy early season wood.

In contrast, at REN, d13C decreased in fast-growing

trees but was constant or increased in slow-growing trees.

Thus, the seasonal pattern is more complex at REN than at

the FLA site and is not simply related to one climatic

factor. However, slow- and fast-growing trees could

represent different rooting depths at this alpine site, which

consists of very stony soils. We cannot rule out that slow-

rowing trees at the REN site experience water stress, but

this was not apparent in the seasonal pattern of fast-

growing trees.

At HAI, d13C generally increased during the year, and

this increase occurred concurrently with a 20% decrease in

precipitation and a 40% decrease in solar radiation

(Table 2). At this site, the effect was most pronounced in

fast-growing trees and was not observable in slow-growing

trees, which is opposite to the observation at REN.

In summary, TRW is correlated with early and late

season precipitation in both the long-term and short-term

time series. However, there was no simple relationship

between d13C and precipitation. There was a significant

correlation with VPD, which is expected to close stomata,

and with radiation, which is expected to open stomata. At

FLA, the ci/ca remained constant during the autumn with

low rainfall but relatively high radiation. A comparison of

the three sites indicates that an interaction with nutrition is

likely. The site with the lowest nitrogen concentrations in

the needles had the lowest d13C values. This result corre-

sponds well with findings from nitrogen fertilization

experiments in which low nitrogen availability resulted in

low d13C in the needles of Pinus sylvestris L. (Hogberg

et al. 1993; Betson et al. 2007), Picea abies Karst.

gc (

mm

olm

-2 s

-1)

0

200

400

600

800

ci/c

a

0.4

0.6

0.8

1.0

δ13C

(‰)

-27.0

-26.0

-25.0

-24.0

-23.0

-22.0

ci/ca

earlywood δ13C

latewood δ13C

Flakaliden

GP

P(µ

mol

m-2 s

-1)

-5

0

5

10

15

20

SP1

SP9

SP3

SP1

SP9

SP3

J F M A M J J A S O N D J F M A M J J A S O N D

2001 2002Date

Fig. 9 Seasonal course of gross

primary productivity (GPP) as

measure of photosynthesis,

canopy conductance (gc), the

ratio of canopy-scale mesophyll

internal and atmospheric CO2

concentration (ci/ca) estimated

based on eddy covariance flux

measurements, and d13C in

wood for the boreal site at

Flakaliden in 2001 and 2002.

Closed symbols of d13C indicate

early wood, open symbolsindicate late wood

Oecologia (2009) 161:729–745 741

123

(Hogberg et al. 1993) and Picea glauca (Moench) Voss

(Livingston et al. 1999). Possible mechanisms include

changes in carboxylation capacity (Livingston et al. 1999)

and an increase in susceptibility to climate-induced water

stress with increasing nitrogen availability (Hogberg et al.

1993; Betson et al. 2007).

The most striking observation is the close correlation

between the early wood of a given year and the late wood

of immediately preceding year, which indicates a carry-

over effect of stored products (Helle and Schleser 2004;

Skomarkova et al. 2006). Thus, our hypothesis was that the

d13C in one year is related to the d13C in the previous year

Table 5 Summary of linear mixed model analyses of the relationships between wood anatomy and carbon isotope ratios and their dependency

on climatic variables

Relationship Tested variablesa Site Variable Interaction site

9 variable

All data d13C vs. anatomy CWA vs. LUM 4.93 130.29*** 21.35***

d13C vs. CWA 6.00* 17.45*** 14.46***

d13C vs. LUM 6.00* 4.62* 64.15***

Early wood CWA vs. LUM 4.03 265.03*** 40.46***

d13C vs. CWA 6.94* 16.19*** 16.25***

d13C vs. LUM 6.94* 29.97*** 24.63***

Late wood CWA vs. LUM 15.60*** 60.51*** 14.86***

d13C vs. CWA 2.95 15.24*** 4.14

d13C vs. LUM 2.95 6.23* 8.27*

TRW vs. climate TRW vs. P-MJJ 3.23 7.38** 16.86***

TRW vs. P-ASO 3.23 0.30 32.60***

TRW vs. T-MJJ 3.23 1.07 1.64

TRW vs. T-ASO 3.23 9.40** 2.85

TRW vs. R-MJJ 3.23 21.67*** 6.04*

TRW vs. R-ASO 3.23 16.51*** 9.61**

TRW vs. VPD-MJJ 3.23 15.01*** 11.09**

TRW vs. VPD-ASO 3.23 18.09*** 5.81

d13C vs. TRW d13C vs.TRW 6.00* 3.19 1.83

Early wood d13C vs. TRW 6.94* 2.75 1.02

Late wood d13C vs. TRW 2.95 4.67* 4.31

TRW vs. growing season TRW vs. length 3.23 11.09*** 9.23**

LUM vs. growing season LUM vs. length 9.43** 594.07*** 0.45

LUM vs. Radiation LUM vs. R-MJJ 9.43** 1089.27*** 1.71

LUM vs. R-ASO 9.43** 1089.11*** 1.12

CWA vs. growing season CWA vs. length 4.93 18.92*** 5.93

d13C vs. growing season d13C vs. length 6.00* 18.92*** 2.14

d13C vs. climate d13C vs. P-MJJ 6.00* \0.01 14.23***

d13C vs. P-ASO 6.00* 2.43 8.55*

d13C vs. T-MJJ 6.00* 0.44 0.31

d13C vs. T-ASO 6.00* 1.08 9.76**

d13C vs. R-MJJ 6.00* 45.01*** 14.58***

d13C vs. R-ASO 6.00* 40.08*** 1.95

d13C vs. VPD-MJJ 6.00* 47.52*** 7.02*

d13C vs. VPD-ASO 6.00* 43.60*** 9.61**

Early vs. late wood preceding year * *** NS

*, **, *** Significant at the 5, 1 and 0.1% level, respectively

Models were fitted by the stepwise inclusion of variables. Listed are the values of likelihood ratio statistics that were applied to assess model

improvement and the statistical significance of the variablesa Variables: TRW, tree-ring width; CWA, cell-wall area; LUM, cell lumen; P, precipitation; T, temperature; R, incoming radiation; VPD, vapor

pressure deficit; length, length of the growing season (days with average temperature [5�C); MJJ, early season (May, June, July); ASO, late

season (August, Spetember, October)

742 Oecologia (2009) 161:729–745

123

through storage products. The use of stored carbohydrates

and lipids for growth in consecutive years has been con-

firmed by Keel et al. (2007) in a permanent 13C-fumigation

experiment of deciduous trees. The accumulation of

reserves involves an accumulation of carbohydrates in the

autumn and a depletion of carbohydrates in the spring. It

also involves an accumulation of lipids during the summer,

which are metabolized in spruce in the autumn (Hoell

1985). However, other tree species, such as Larix, may

accumulate lipids during the winter (Sudachkova et al.

2004). The effect of storage has been demonstrated in

deciduous trees (Helle and Schleser 2004) which have a

larger number of wood parenchyma cells than conifers

(Schweingruber et al. 2007). One would not expect major

storage pools in conifers, which have only a few ray cells.

However, Hoell (1985) did demonstrate the turnover of

soluble carbohydrates in conifers, and Von Felten et al.

(2007) estimated that 42% of new wood in Larix originates

from storage products.

Direct proof that stored carbohydrates of one year con-

tribute to the growth of the next year was presented for

conifers by Hansen and Beck (1994), who pulse-labeled

young Pinus sylvestris trees with 14C in the field and

demonstrated the incorporation of 14C into wood of the

following year. The process was biochemically further

complicated by the fact that starch was transformed into

various soluble carbohydrates, such as maltotriose, gal-

actose, arabinose and rhamnose in the winter as a mecha-

nism of cold hardiness (Hansen and Beck 1994). Thus, the

breakdown of starch in wood results in the production of

triose-phosphates, which results again in a 13C depletion

step (Gleixner et al. 1998). Obviously, it is also not pos-

sible to demonstrate fluxes of 13C in retrospect using tree

rings and ray cells.

Gessler et al. (2008) measured d13C real time in the

phloem sap of Rhizinus and found an increase as well as a

decrease in d13C in the phloem sap compared to the

immediate photosynthetic product, depending on whether

the phloem sugars were derived from sucrose or from

starch or from sucrose passing through the cell metabolic

pathways, including the formation of triose-phosphates. No

fractionation was found during phloem transport. However,

this experiment did not include the formation of wood in a

seasonal climate, which may be supported by carbohy-

drates from ray cells in the wood and not involve phloem

transport. This situation is different from that of continually

fast-growing trees (Pinus radiata: Walcroft et al. 1997;

Eucalyptus globulus: Pate and Arthur 1998) where a cor-

relation between the phloem sap isotope signal and d13C in

wood has been demonstrated. During the loading of storage

cells, carbohydrates move from the phloem via the cam-

bium and the xylem initials into the ray cells of existing

wood. If wood formation has priority over storage, the

storage products would be expected to become relatively

enriched in 13C (increasing d13C). However, if storage and

cold hardiness has priority, then d13C could also be

depleted (decreasing d13C). In contrast, during carbohy-

drate mobilization in the spring, carbohydrates move from

ray cells into the cambium without entering the phloem. It

would be the mixture of the enriched mobilized carbohy-

drates and the phloem-derived carbohydrates from spring

photosynthesis which determines the d13C of wood. An

additional important step of discrimination takes place

during wood respiration due to the breakdown of carbo-

hydrates into triose-phosphates. However, respiratory CO2

is 13C enriched (Brandes et al. 2006), which does not

explain the enrichment of d13C in stems (Damesin and

Lelarge 2003).

In our study, a stepwise change between late and early

wood of consecutive years was observed, but this stepwise

change was not always in the same direction, as was pre-

dicted by Helle and Schleser (2004). We observed all

possible permutations of alternative responses depending

on the site, species and climate. Thus, the linkage between

the d13C of wood and climate is not constant. There are

many ways which link d13C in wood, storage and climate

and which govern the relation between tree ring width and

d13C. Thus, based on our results, we conclude that tree-ring

width is not consistently related to d13C in seasonal

climates, and both are only indirectly linked to climate.

Tre

e rin

g w

idth

(m

m)

0.5

1.0

1.5

2.0

2.5

3.0

3.5

FLA

HAI

REN

Picea abies

Year1999 2000 2001 2002 20031999 2000 2001 2002 2003

δ13C

(‰)

-26.5

-26.0

-25.5

-25.0

-24.5

-24.0

-23.5

FLA

HAI

REN

Fig. 10 Inter-annual variations

in absolute values of d13C in

tree rings of spruce (all three

studied sites) during the last 5

years (1999–2003). Verticallines show the standard

deviations

Oecologia (2009) 161:729–745 743

123

Conclusion

The carry-over effect, based on correlations between the

d13C of early wood and the d13C of late wood of the

immediately preceding year is significant in conifers.

However, depending on sites and specific meteorological

conditions, other effects, such as whole-wood composition,

the isotope signal of storage products or a dry autumn

leading to dry spring result in additional variation.

The use of a 5�C average temperature as a proxy for the

start of the growing season appears to be robust for boreal,

temperate and Mediterranean sites. The correlations in this

study are insensitive to this proxy, but more elaborate

phasing studies are needed in the future.

Acknowledgments This work was supported by Alexander von

Humboldt (Research Award 2003 for E. Vaganov) and the Russian

Foundation of Basic Research (RFBR-05-04-48069). We thank

Alessandro Cescatti, Leonardo Montagnani, Stefano Minerbi and

Claudio Mutinelli for providing the climate and nitrogen data for

Renon, Sune Linder for dendrometer data, and Anders Lindroth for

eddy flux data of the Flakaliden site. We thank Gerd Gleixner

for discussion of this manuscript. We also like to thank Annett

Boerner for the artwork and Jens Schumacher for advice on statistical

analyses.

Open Access This article is distributed under the terms of the

Creative Commons Attribution Noncommercial License which per-

mits any noncommercial use, distribution, and reproduction in any

medium, provided the original author(s) and source are credited.

References

Adams HD, Kolb TE (2004) Drought responses of conifers in ecotone

forests of northern Arizona: tree ring growth and leaf d13C.

Oecologia 140:217–225

Bates D, Sakar D (2006) LME4: linear mixed-effects models using S4

classes. R package version 0.995-2. Available at:

http://www.r-project.org

Betson NR, Johannisson C, Lofvenius MO, Grip H, Granstrom A,

Hogberg P (2007) Variation in the delta C-13 of foliage of Pinussylvestris L. in relation to climate and additions of nitrogen:

analysis of a 32-year chronology. Glob Chang Biol 13:2317–2328

Brandes E, Kodama N, Whittaker K, Weston C, Rennenberg H, Keitel

C, Adams MA, Gessler A (2006) Short-term variation in the

isotopic composition of organic matter allocation from the leaves

to the stem of Pinus sylvestris: effects of photosynthetic and

postphotosynthetic carbon isotope fractionation. Glob Chang

Biol 12:1922–1939

Briffa KR, Schweingruber FH, Jones PD, Osborn TJ, Shiyatov SG,

Vaganov EA (1998) Reduced sensitivity of recent tree growth to

temperature at high northern latitudes. Nature 391:678–682

Brugnoli E, Farquhar GD (2000) Photosynthetic fractionation of

carbon isotopes. In: Leegood RC, Sharkey TD, von Caemmerer

T (eds) Advances in photosynthesis: physiology and metabolism,

vol 9. Springer, Berlin Heidelberg New York, pp 399–434

Cescatti A, Marcolla B (2004) Drag coefficient and turbulence

intensity in conifer canopies. Agric For Meteorol 121:197–206

Cook ER, Kairiuktis LA (eds) (1990) Methods of dendrochronology.

Application in the environmental sciences. Kluwer, Dordrecht

Cook ER, Peters K (1981) The smoothing spline: a new approach to

standardizing forest interior tree-ring width series for dendrocli-

matic studies. Tree Ring Bull 41:45–53

Cook ER, Briffa KR, Shiyatov SG, Mazepa VS (1990) Tree-ring

standardization and growth-trend estimation. In: Cook ER,

Kairiukstis LA (eds) Methods of dendrochronology. Application

in the environmental sciences. Kluwer, Dordrecht, pp 104–123

Damesin C, Lelarge C (2003) Carbon isotope composition of current-

year shoots from Fagus sylvatica in relation to growth,

respiration and use of reserves. Plant Cell Environ 26:207–219

Duquesnay A, Breda N, Stievenard M, Duponuey JL (1998) Changes

of tree ring delta C-13 and water-use efficiency of beech (Fagussylvatica L.) in north-eastern France during the past century.

Plant Cell Environ 21:565–572

Eglin T, Maunoury-Danger F, Fresnau C, Lelarge C, Pollet B,

Lapierre C, Francois C, Damesin C (2008) Biochemical

composition is not the main factor influencing variability in

carbon isotope composition of tree rings. Tree Physiol 28:1619–

1629

Farquhar GD, Ehleringer JR, Hubik KJ (1989) Carbon isotope

discrimination and photosynthesis. Annu Rev Plant Physiol Plant

Mol Biol 40:503–537

Ferrio JP, Florit A, Vega A, Serrano L, Voltas J (2003) Delta c-13 and

tree-ring width reflect drought responses in Quercus ilex and

Pinus halepensis. Oecologia 137:512–518

Francey RJ, Farquhar GD (1982) An explanation of C-13/C-12

variations in tree rings. Nature 297:28–31

Fritts HC (1966) Growth-rings of trees: their correlation with climate.

Science 154:873–879

Fritts HC (1976) Tree-rings and climate. Academic Press, London

Furst GG (1979) Methods of anatomical and histochemical research

of plant tissue (in Russian). Nauka, Moscow

Gessler A, Tcherkez G, Peuke AD, Ghashghaie J, Farquhar GD

(2008) Experimental evidence for diel variations of the carbon

isotope composition in leaf, stem and ploem sap organic matter

in Ricinus communis. Plant Cell Environ 31:941–953

Gleixner G, Scrimgeour C, Schmidt HL, Viola R (1998) Stable

isotope distribution in major metabolites of source and sink

organs of Solanum tuberosum L.: a powerful tool in the study of

metabolic partitioning in intact plants. Planta 207:241–245

Gonzales IG, Eckstein D (2003) Climate signal of earlywood vessels

of oak in a maritime climate. Tree Physiol 23:497–504

Hansen J, Beck E (1994) Seasonal changes in the utilization and

turnover of assimilation products in 8-year-old Scots pine (Pinussylvestris L.) trees. Trees 8:172–182

Harlow BA, Marshall JD, Robinson AP (2006) A multi-species

comparison of d13C from whole wood, extractive-free wood and

holocellulose. Tree Physiol 26:767–774

Helle G, Schleser GH (2004) Beyond CO2-fixation by Rubisco—an

interpretation of 13C/12C variations in tree rings from novel intra-

seasonal studies on broad-leaf trees. Plant Cell Environ 27:367–

380

Hemming D, Fritts H, Leavitt SW, Wright W, Long A, Shashkin A

(2001) Modelling tree-ring d13C. Dendrochronologia 19:23–38

Hill SA, Waterhouse JS, Field EM, Switsur VR, Aprees T (1995)

Rapid recycling of triose phosphates in oak stem tissue. Plant

Cell Environ 18:931–936

Hoell W (1985) Seasonal fluctuations of reserve materials in the

trunkwood of spuce (Picea abies (L.) Karst.). J Plant Physiol

117:355–362

Hogberg P, Johannisson C, Hallgren JE (1993) Studies of C-13 in the

foliage reveal interactions between nutrients and water in forest

fertilization experiments. Plant Soil 152:207–214

744 Oecologia (2009) 161:729–745

123

Holmes RL (1983) Computer-assisted quality control in tree-ring

dating and measurement. Tree Ring Bull 44:69–75

Holmes RL (1992) Program COFECHA: Version 3. The University of

Arizona, Tucson

Hubbard RM, Stiller V, Ryan MG, Sperry JS (2001) Stomatal

conductance and photosynthesis vary linearly with plant hydraulic

conductance in ponderosa pine. Plant Cell Environ 24:113–121

Hughes MK, Vaganov EA, Shiyatov SG, Touchan R, Funkhouser G

(1999) Twentieth-century summer warmth in northern Yakutia

in a 600-year context. Holocene 9:603–608

Jones PD, Briffa KR (1992) Global surface air temperature variations

during the twentieth century: part 1, spatial, temporal and

seasonal details. Holocene 2:165–179

Kagawa A, Naito D, Sugimoto A, Maximov TC (2003) Effects of

spatial and temporal variability in soil moisture on widths and

d13C values of eastern Siberian tree rings. J Geophys Res

108:4500

Kagawa A, Sugimoto A, Maximov TC (2006a) (CO2)-C13 pulse-

labelling of photoassimilates reveals carbon allocation within

and between tree rings. Plant Cell Environ 29:1571–1584

Kagawa A, Sugimoto A, Maximov TC (2006b) Seasonal course of

translocation, storage and remobilization of C-13 pulse-labeled

photoassimilate in naturally growing Larix gmelinii saplings.

New Phytol 171:793–804

Keel SG, Siegwolf RTW, Jaggi M, Koerner C (2007) Rapid mixing

between old and new C pools in the canopy of mature forest

trees. Plant Cell Environ 30:963–972

Kirdyanov A, Hughes M, Vaganov E, Schweingruber F, Silkin P

(2003) The importance of early summer temperature and date of

snow melt for tree growth in the Siberian Subarctic. Trees

17:61–69

Knohl A, Schulze ED, Kolle O, Buchmann N (2003) Large carbon

uptake by an unmanaged 250-year-old deciduous forest in

Central Germany. Agric For Meteorol 118:151–167

Lambers H, Chapin FSIII, Pons TL (1998) Plant physiological

ecology. Springer, New York, p 540

Leavitt SW (1993) Environmental information from 13C/12C ratios of

wood. Geophys Monogr 78:325–331

Leikola M (1969) The influence of environmental factors on the

diameter growth of young trees. Acta For Fenn 92:1–44

Linder S (1995) Foliar analysis for detecting and correcting nutrient

imbalances in Norway spruce. Ecol Bull (Copenhagen) 44:178–

190

Livingston NJ, Guy RD, Sun ZJ, Ethier GJ (1999) The effects of

nitrogen stress on the stable carbon isotope composition, produc-

tivity and water use efficiency of white spruce (Picea glauca(Moench) Voss) seedlings. Plant Cell Environ 22:281–289

Lundmark T, Bergh J, Strand M, Koppel A (1998) Seasonal variation

of maximum photochemical efficiency in boreal Norway spruce

stands. Trees 13:63–67

McCarroll D, Loader NJ (2004) Stable isotopes in tree rings. Quat Sci

Rev 23:771–801

McNulty SG, Swank WT (1995) Wood delta-C-13 as a measure of

annual basal area growth and soil-water stress in a pinus-strobus

forest. Ecology 76:1581–1586

Mikola P (1962) Temperature and tree growth near the northern

timberline. In: Kozlowski TT (ed) Tree growth. Ronald, New

York, pp 265–287

Monserud RA, Marshall JD (2001) Time-series analysis of delta C-13

from tree rings I. Time trends and autocorrelation. Tree Physiol

21:1087–1102

Munro MAR, Brown PM, Hughes MK, Garcia EMR (1996) Image

analysis of tracheid dimensions for dendrochronological use. In:

Dean JS, Meko DM, Swetnam TW (eds) Tree rings, environment

and humanity radiocarbon. Birkhauser, Basel, pp 843–853

Oren R, Schulze ED (1989) Nutritional disharmony and forest

decline: a conceptual model. Ecol Stud 77:425–443

Pate J, Arthur D (1998) d13C analysis of phloem sap: novel means of

evaluating seasonal water stress and interpreting carbon isotope

signatures of foliage and trunk wood of Eucalyptus globulus.

Oecologia 117:301–311

Rinn F (1996) Tsap V 3.6 reference manual: computer program for

tree-ring analysis and presentation. RINNTECH, Heidelberg

Scartazza A, Mata K, Matteucci G, Yakir D, Moscatello S, Brugnoli E

(2004) Comparisons of d13C of photosynthetic products and

ecosystem respiratory CO2 and their responses to seasonal

climate variability. Oecologia 140:340–351

Schulze B, Wirth C, Linke P, Brand WA, Kuhlmann I, Horna de

Zimmermann V, Schulze ED (2004) Laser-Ablation-Combus-

tion-GC-IRMS—a new method for online analysis of intra-

annual variation of d13C in tree ring. Tree Physiol 24:1193–1201

Schulze ED, Beck E, Muller-Hohenstein K (2006) Plant ecology.

Springer, Heidelberg, p 846

Schweingruber FH (1988) Tree ring: basics and applications of

dendrochronology. Reidel Publishing, Dordrecht

Schweingruber FH (1996) Tree rings and environment. Dendroecol-

ogy. Paul Haupt, Bern

Schweingruber FH, Boerner A, Schulze ED (2007) Atlas of woody

plant stems. Springer, Heidelberg

Shiyatov SG (1986) Dendrochronology of upper timberline in ural

mountains (in Russian). Nauka, Moscow

Skomarkova MV, Vaganov EA, Mund M, Knohl A, Linke P, Boerner

A, Schulze ED (2006) Inter-annual and seasonal variability of

radial growth, wood density and carbon isotope ratios in tree

rings of beech (Fagus sylvatica) growing in Germany and Italy.

Trees 20(5):571–586

Sudachkova NE, Milyutina IL, Romanova LI, Semenova GP (2004)

The annual dynamics of reserve compounds and hydrolytic

enzyme activity in the tissues of Pinus sylvestris L and Larixsibirica Lebed. Eurasian J For Res 7:1–10

Vaganov EA (1990) The traheidogram method in tree-ring analysis

and its application. In: Cook ER, Kairiuktis LA (eds) Methods of

dendrochronology. Applications in the environmental sciences.

Kluwer, Dordrecht, pp 63–75

Vaganov EA, Shashkin AV, Sviderskaya IV, Vysotskaya LG (1985)

Histometric analysis of woody plant growth (in Russian). Nauka,

Novosibirsk

Vaganov EA, Shiyatov SG, Mazepa VS (1996) Dendroclimatic study

in Ural-Siberian Subarctic (in Russian). Nauka, Novosibirsk

Vaganov EA, Hughes MK, Shashkin EA (2006) Growth dynamics of

conifer tree rings: images of past and future environments.

Springer, Heidelberg

Von Felten S, Hattenschwiler S, Saurer M, Siegwolf R (2007) Carbon

allocation in shoots of alpine treeline conifers in a CO2 enriched

environment. Trees 21:283–294

Walcroft AS, Silvester WB, Whitehead D, Kelliher FM (1997)

Seasonal changes in stable carbon isotope ratios within annual

rings of Pinus radiata reflect environmental regulation of growth

processes. Aust J Plant Physiol 24:57–68

Walter H, Lieth H (1967) Klimadiagramm Weltatlas. Fischer, Jena

Wilson AT, Grinsted WA (1977) 12C/13C in cellulose and lignin as

paleothermometers. Nature 265:133–135

Wimmer RP (2002) Wood anatomical features in tree-rings as

an indicators of environmental changes. Dendrochronologia

20:21–36

Oecologia (2009) 161:729–745 745

123